A Business Intelligence tool applied to a Manufacturing simulation environment Authors: Alfredo Elia ( Advanced Engineering Solutions ), Guido

A Business Intelligence tool applied to a Manufacturing simulation environment Authors: Alfredo Elia ( Advanced Engineering Solutions ), Guido Vindrola ( Amet Italy ).

Topics Simulation of a manufacturing process in non deterministic conditions, by using DOE design. Howto analyzedoe multi dimentionaldata space with HiQube. Process Automation to grab results data and populate HiQube database. 2

ISSUE The production of a particular is not stable due to the influence of non controllable factors. The main aim of this work is to simulate forming process in non deterministic conditions and to define and control the KPIs. 3

Analysis Pipeline 4

NUMERICAL MODEL HyperForm is used to create the finite element model, to assign the boundary conditions and to build the LS-Dyna input deck. The multi stage forming process is simulated with an appropriate material model (*MAT_37*MAT_37 in LS- Dyna) able to take into account the anisotropic characteristic. 5

Nominal Run Analysis FLD diagram 6

Nominal Run Analysis Percent Thickness Reduction 7

Nominal Run Analysis Post Processing considerations It is evident a uniform thickness achieved after forming, except for a local portion of the sheet affected by thinning (in any case under 30%) and a local compression zone along the lateral wall of the component. 8

DOE A DOE design is conducted to analyze the relationship between the Punch Force and the process variables like: Anisotropy Friction coefficient Thickness Yelding Point Young Module 9

DOE A 2 k design is chosen with a single replication, one center point and axial points with α=1,596. It is important to remind that, according to this design, it is possible to build a response surface, by using an additive model as follows: Y= β₀+ i β i x i + i j β ij x i x j + i β ii x i 2 1 st Order Full Quadratic 10

Corner Points 1 (A,B,C)=(1,1,1) FACTOR C -1 1 FACTOR B FACTOR A 1 (A,B,C)=(1,1,- 1) 11

Center and Axial Points (A,B,C)=(1.596,0,0) 1.596 0 FACTOR C 0 1.596 FACTOR A FACTOR B Center Point (A,B,C)=(0,0,0) -1.596 0 1.596 12

Uncoded values for factors levels FACTORS Lev. -1,596 Lev. -1 Lev. 0 Lev. 1 Lev. 1,596 Young Module [MPa] 2,050E+02 2,063E+02 2,085E+02 2,107E+02 2,120E+02 Thickness [mm] 1,400E+00 1,437E+00 1,500E+00 1,563E+00 1,600E+00 Yielding Point [MPa] 3,078E+02 3,206E+02 3,420E+02 3,634E+02 3,762E+02 Friction [adim] 1,100E-01 1,175E-01 1,300E-01 1,425E-01 1,500E-01 Anisotropy [adim] 1,000E+00 1,037E+00 1,100E+00 1,163E+00 1,200E+00 13

Run Table with Punch Force values RUN ID Anisotropy Friction Thickness Yielding Point Young Module Punch Force 1-1 -1-1 -1-1 1180350,00 2-1 -1-1 -1 1 1272680,00 3-1 -1-1 1-1 1157930,00 4-1 -1-1 1 1 1276080,00 5-1 -1 1-1 -1 1180340,00 6-1 -1 1-1 1 1276080,00 7-1 -1 1 1-1 1180340,00 8-1 -1 1 1 1 1276080,00 9-1 1-1 -1-1 1183550,00 10-1 1-1 -1 1 1275900,00 11-1 1-1 1-1 1183550,00 12-1 1-1 1 1 1275900,00 13-1 1 1-1 -1 1183550,00 14-1 1 1-1 1 1275900,00 15-1 1 1 1-1 1183550,00 16-1 1 1 1 1 1275900,00 17 1-1 -1-1 -1 1175850,00 18 1-1 -1-1 1 1267250,00 19 1-1 -1 1-1 1175850,00 20 1-1 -1 1 1 1267250,00 21 1-1 1-1 -1 1175850,00 22 1-1 1-1 1 1267250,00 23 1-1 1 1-1 1175850,00 24 1-1 1 1 1 1267250,00 Corner Points 14

Run Table with Punch Force values RUN ID Anisotropy Friction Thickness Yielding Point Young Module Punch Force 22 1-1 1-1 1 1267250,00 23 1-1 1 1-1 1175850,00 24 1-1 1 1 1 1267250,00 25 1 1-1 -1-1 1180050,00 26 1 1-1 -1 1 1271340,00 27 1 1-1 1-1 1180050,00 28 1 1-1 1 1 1271340,00 29 1 1 1-1 -1 1180050,00 30 1 1 1-1 1 1271340,00 31 1 1 1 1-1 1180050,00 32 1 1 1 1 1 1271340,00 41-1,596 0 0 0 0 1229200,00 42 1,596 0 0 0 0 1218830,00 39 0-1,596 0 0 0 1222520,00 40 0 1,596 0 0 0 1228170,00 37 0 0 --1,596 0 0 1224180,00 38 0 0 1,596 0 0 1224180,00 35 0 0 0-1,596 0 1224180,00 36 0 0 0 1,596 0 1224180,00 33 0 0 0 0-1,596 1150370,00 34 0 0 0 0 1,596 1299070,00 43 0 0 0 0 0 1224180,00 Corner Points Axial Points Center Point 15

Response Surface Coefficients Term Coef P-value Constant 1223675 0 Young Module 46892 0 Yielding point -513 0,401 Thickness 696 0,258 Friction 2159 0,002 Anisotropy -2056 0,002 Primary Effects Young Module*Young Module 450 0,661 Yielding point*yielding point 238 0,816 Thickness*Thickness 238 0,816 Friction*Friction 696 0,499 Anisotropy*Anisotropy 174 0,865 Young Module*Yielding point 807 0,224 Young Module*Thickness -594 0,367 Young Module*Friction -1031 0,124 Young Module*Anisotropy -1269 0,062 Yielding point*thickness 594 0,367 Yielding point*friction 594 0,367 Yielding point*anisotropy 594 0,367 Thickness*Friction -806 0,224 Thickness*Anisotropy -806 0,224 Friction*Anisotropy -149 0,820 Secondary Effects 16

Simulation Ls-Dyna has been used as solver to compute 43 runs generated from DOE. 17

PostProcessing The advantage of Process Automation All results data presented above have been extracted from the whole results data by using Process Automation tools integrated in Hyperworks framework. 18

PostProcessing The advantage of Process Automation A Process Manager integrated in HyperView has been developed, in order to lead user to extract Process Responses (i.e. Punch Force) and to define Control Ponits for thickness variation monitoring. 19

Post Process in 3 steps 20

Define Result Structure 21

Define Selection Spheres 22

FLD Setup 23

Process Manager Output Process Manager produces a CSV file (Comma Separated File) to store the results for each DOE run, in terms of: Punch force FLD Image Thickness Map Image Control Points thickness 24

Engineering Intelligence ROW DATA We adopted DSS (Decision Support Systems) methodology to define Key Performance Indicators (KPYs) of stamping process, what is referred to as an Engineering Intelligence approach. 25

HiQube HiQube is a DSS tool based on a propietary OLAP (On-Line Analytical Process) Database which is at the main time: Hierarchical Cartesian Relational 26

HiQube A simple panel is built in order to browse results in main two different ways: Analyze Primary effects on Punch Force Response Analyze two factors interactions on the same response. 27

Primary effects Plot 28

Two Factors Interactions Plot 29

Drill Through to plot FLD 30

Control Points A control point can be considered as a spherical domain of selection of some elements of the blank. The value of thickness assigned to a control point is the average value of all elements inside the selection sphere associated to that control point. 31

Thickness Map with Control Points A set of control points is defined, in order to take track of percent thickness deviation from the value related to a specific run simulation. An icon legend is used to offer an immediate feedback to the user: Out of upper bound In control Out of lower bound 32

Drill Through Thickness Control Points 33

Conclusions It has been shown how HiQube can be used in a process automation flow, in order to perform a DOE analysis of a manufacturing process. This application takes advantage by the fact that the used tools are all integrated in one CAE solutions suite: Hyperworks The developed application can be considered as a Proof Of Concept for the capablities monitoring of a simulated process, which of course could be extended in other application fields. 34